Towards Generalizable Robotic Data Flywheel: High-Dimensional Factorization and Composition

TL;DR

This paper introduces F-ACIL, a framework that decomposes robotic data into structured factors to improve data efficiency and generalization, achieving significant performance gains with fewer demonstrations.

Contribution

F-ACIL is a novel factor-aware compositional learning framework that enables structured data factorization and promotes generalization in robotic models.

Findings

Achieves over 45% performance improvement

Uses 5-10 times fewer demonstrations

Demonstrates effective structured factorization

Abstract

The lack of sufficiently diverse data, coupled with limited data efficiency, remains a major bottleneck for generalist robotic models, yet systematic strategies for collecting and curating such data are not fully explored. Task diversity arises from implicit factors that are sparsely distributed across multiple dimensions and are difficult to define explicitly. To address this challenge, we propose F-ACIL, a heuristic factor-aware compositional iterative learning framework that enables structured data factorization and promotes compositional generalization. F-ACIL decomposes the data distribution into structured factor spaces such as object, action, and environment. Based on the factorized formulation, we develop a factor-wise data collection and an iterative training paradigm that promotes compositional generalization over the high-dimensional factor space, leading to more effective utilization of real-world robotic demonstrations. With extensive real-world experiments, we show that F-ACIL can achieve more than 45% performance gains with 5-10$\times$ fewer demonstrations comparing to that of which without the strategy. The results suggest that structured factorization offers a practical pathway toward efficient compositional generalization in real-world robotic learning. We believe F-ACIL can inspire more systematic research on building generalizable robotic data flywheel strategies. More demonstrations can be found at: https://f-acil.github.io/